Effects of the genetic targeting of apelinergic system

In the study published a few months before the ending of this thesis, Jaiprasart et al. have been able to follow the growth of antiangiogenic-resistant tumors³³⁸. In their study, authors have successfully identified the group of mice bearing tumors that have gained adaptive resistance to antiangiogenic drugs, such as bevacizumab and sorafenib. Resistant tumors (RT) resumed angiogenesis and cell proliferation despite the continuation of applied treatment. Finally, they found that, amongst other genes, APJ and APLN were among the top enriched genes within the stroma of resistant tumors and demonstrated that regulation of these genes has a strong connection to resistance to antiangiogenic therapy.

DISCUSSION

125

Before the beginning of this project, a research group from our department has already published the study on successful targeting of pulmonary endothelial cells by using adeno-associated viral vectors (AAV)²⁷⁷. The authors showed that a selected capsid variant allows AAV entry exclusively into the endothelium of the pulmonary vasculature after intravenous injection and therefore assures efficient delivery of the gene of interest to the selected entity.

This finding was alluring for the purpose of further investigation in this study, since AAV vectors could be utilized in order to genetically “impose” the stable expression of Apelin to the murine lung endothelium. Overexpressing Apelin was expected to give valuable information on its effect on tumor cell homing and, finally, insights into metastatic onset. Having tested the intravenous lung metastasis model, technically complex cloning of Apelin inside AAV was performed. Furthermore, to include the effect coming from the inhibition of apelinergic signaling into the experimental setup, the gene encoding for F13A was cloned into AAV vector.

AAV vectors additionally contained a gene encoding for firefly luciferase, which could be easily used as a reporter of a stable AAV integration. Results from in vitro experiments and sequencing have confirmed successful cloning. Additionally, the goal was to confirm the specificity of AAV in used experimental conditions. Based on the results from bioluminescent imaging and already published findings, successful and specific targeting of lungs was achieved.

Contrary to all expectations, the main in vivo experiment showed entirely different results.

Substantially decreased tumor cell engraftment revealed a low metastatic burden that was present within all study groups. Due to these findings, no antiangiogenic effect could be detected in comparison to any vehicle control group (PBS subgroups). Accordingly, no clear conclusion on apelinergic system’s influence on metastasis onset could be made, since some of the observed lung samples were even metastasis-free. These unusual findings were further confirmed by analyzing the gene expression of eGFP, expecting it to correlate with the number of tumor cells present in the lungs. It is worth noticing that the infiltration and engraftment of tumor cells were not as optimal as it was observed in the preliminary intravenous lung metastasis model.

Yet, the reason for low metastatic colonization was not clear, because the same cells and the matched mouse subjects (in regard to sex and age) were used in both experiments. Although their substrains were equal (6J) and therefore could not be considered as different, C57BL mice from the first and last two intravenous experiments were obtained from two sources (inbred and from commercial supplier Janvier, respectively). Therefore, one can speculate that the animal

DISCUSSION

126

immune system or microbiome might have been changed and consequently had an impact on cell engraftment after tumor cell injection. Although differences in the genetics were likely to exist, this scenario was excluded, since all inbred lines are regularly re-constituted in the facility and at Janvier to minimize the possibility of genetic drift.

The limiting factor of the first conducted AAV experiment was the deficiency of a control group with no AAV treatment, since in this experimental setup, all groups were treated with AAV vectors (including AAV empty vector group). As a possible explanation for the unusual results, it was hypothesized that the applied vectors might somehow manifest an immune-mediated effect on tumor cell engraftment. Therefore, it was thought that the immune system, triggered by the presence of AAV vectors in lung endothelium, simply eliminates the tumor cells and that these events occur randomly, as shown by the variations of the lung metastatic burden between different groups. Ambiguous results from the previous experiments led to the assumption that the immune system, activated by presence of AAV vectors might influence the cell engraftment.

From personal communication with Prof. Dr. Jakob Körbelin, who designed the lung-EC-specific vector, it was known that no observable influence on tumor colonization due to AAV injection was previously noted. However, these claims must be taken with caution, since different mouse models have been tested and therefore, an influence of the immune system in this experimental system cannot be excluded.

Since the first conducted experiment was lacking the no-AAV control group, the main mouse experiment was repeated by including an additional control group (no-AAV, vector-free mice).

However, even with the improved experimental setup, results similar to the ones from the first experiment were obtained. On the other hand, it is worth noticing that the engraftment was moderately higher in comparison with the first experiment, but the high variability between all groups remained the same. No clear conclusion on the effect of any group parameter (agonistic or antagonistic effect of Apelin/F13A or antiangiogenic therapy) on metastatic burden could have been made after macroscopic observation.

One may, however, question the choice of mouse strain. The potential lung metastasis model in immunodeficient NSG mice would be more appropriate, since it is expected that mice without immune system would be more likely to have had higher metastatic colonization in comparison to those with a fully functional immune system. Indeed, several metastatic models have already been described^339,340. However, it must be pointed out that the main research goal of this study was to investigate the role of immune system and apelinergic signaling on tumor

DISCUSSION

127

cell homing and metastatic onset, as well as on antiangiogenic treatment. For that reason, C57BL6 mice, with an intact immune system, were chosen as a model. Poor engraftment that was shown in both experiments was the limiting factor of this study. Although tumor engraftment is highly dependent on the mouse strain, this part of the experimental setup remained unchanged. Certain tumors, such as prostate carcinoma, do not engraft well in some mouse strains, and therefore high engraftment rate is unlikely to be achieved³⁴¹. In this study, the engraftment abilities of MC38 cell line that has been used in main experiment, can also be speculated. On the one hand, metastatic potential of MC38 cells in intravenous models has already been evaluated by Kryczek et al.³⁴². Additionally, in preliminary intravenous lung metastasis model, MC38 cells displayed the highest capacity to generate metastasis. However, that capacity was not retained using passaged cells for the main experiments. We could also assume that the viability of applied tumor cells could have been the reason for low metastatic burden, since the effective implementation of this model requires a higher degree of technical precision. On the other hand, in some cases, certain subpopulations of the cancer cells have been shown to be more efficient at generating tumors in mice compared to other cell subpopulations³⁴³. Therefore, it might have happened that during the passaging, cells have lost their engraftment potential. Taken together, the likelihood that MC38 cells suffered a phenotypic change cannot be excluded. Most of the limitations of this model could have been overcome by other approaches, such as activating the oncogenes like c‐Neu oncogene, driven by a mouse mammary tumor virus promoter³⁴⁴, which has been proven to be of use in the study of Uribesalgo et al.¹⁵⁰. That way, more reproducible results could have been obtained.

Immunohistochemical (IHC) analysis of the lung tissues revealed to some extent a similar infiltration rate of the immune cells in all observed groups. In the majority of analyzed groups, high variability in CD3, as well as in CD11b area, was observed. In case of myeloid cell lineage (CD11b) it could be seen that their infiltration was lower in F13A-DC101 group, compared to all other groups. Yet, statistically significant difference was observed only when compared to DC101-treated no-vector group. Interestingly, the myeloid cell infiltration was shown to be decreased in apelin-KO mouse model heart tissue in study of Tatin et al.³⁴⁵. In the experimental settings used in this project, it can be assumed that presence of F13A was the reason for a decreased percentage of CD11b positive area. However, these findings need to be confirmed in a separate experimental setup.

On the other hand, T-cell marker (CD3) positive area was shown to be the highest in DC101-treated Apelin group. The increase in CD3 coverage was contrary to the results of

DISCUSSION

128

Yamazaki et al., who showed that infiltration of T cells is significantly inhibited by Apelin secretion in another pathological condition³⁴⁶. Since mice from analyzed Apelin group in this experimental system were treated with antiangiogenic therapy, results may indicate that DC101 abrogated the effect of Apelin, and that this could have been sufficient to induce infiltration of CD3 cells. On the other hand, these findings were partially in agreement with the research of Tatin et al. The authors have analyzed distribution of immune cell populations in the heart of apelin-KO mice compared with control mice and did not observe any changes in the ratio of CD3-positive T lymphocytes³⁴⁵.

In terms of microvessel density, moderate differences in CD31 tumor coverage between different groups could be observed. As previously mentioned, it was speculated that the apelinergic system has a central role in tumor endothelium. Compared to DC101-treated no-vector control, both Apelin groups displayed significantly higher CD31 tumor coverage. This may indicate that Apelin, secreted from pulmonary endothelial cells, additionally increased tumor microvessel density. However, if we compare two Apelin subgroups (control and DC101-treated), no changes in MVD could be observed, suggesting the possibility that DC101 effect could have been inhibited by the presence of Apelin. Similar situation could be seen in F13A group, since there was no difference between control and treated subgroup. On the other hand, differences in terms of MVD become more evident if we compare Apelin and F13A groups. Significantly decreased MVD in F13A group can be explained by antagonism of APJ by F13A. These findings were in accordance with the recently published study¹⁵⁰. Still, it is of high importance to confirm these findings by additional experimental setup, since high variability in analysis of MVD can present a potential bias. Therefore, these results should be taken with caution.

Despite the proven specificity of the AAV vectors toward mouse lung endothelium, one can speculate that Apelin or F13A level might not have been secreted by pulmonary endothelial cells in sufficient quantity to influence tumor metastasis onset. Therefore, Apelin protein level in all murine lungs was quantified. Accordingly, the analysis revealed low Apelin protein level in all processed lung tissues and no differences between control and AAV-Apelin groups could have been observed. However, concentration of F13A was higher compared to all other groups.

One can imply that the low Apelin protein level might have been the reason for variations of metastatic burden, but still affected MVD. Yet, antiangiogenic treatment did not have an expected effect in treated groups, probably due to a random effect in cell engraftment that has been present. It can be concluded that these experimental conditions were not optimal for the

DISCUSSION

129

complexity of the desired model, indicating that the next model will require substantial reassessment in terms of higher controllability and reproducibility.

In the beginning of this project, studies with similar hypotheses were conducted from separate working groups. In the experimental system using oncogene-driven spontaneously induced mammary cancer, Uribesalgo et al. have proven that loss of Apelin reduced tumor angiogenesis and impaired tumor growth¹⁵⁰. As an advantage, the authors used a previously established model of Apelin-KO mice, which was a more controllable system in comparison to the one used in this thesis. Their system allowed an insight into the effect of Apelin produced from the tumor cells and tumor-microenvironment-derived Apelin. Finally, authors managed to prove that the depletion of Apelin from any of the mentioned compartments is sufficient to reduce tumor growth. Apelin depletion was also proven to be the key factor for tumor microenvironment remodeling, resulting in improved vessel leakiness and a decrease in the infiltration of immune-suppressive subset of neutrophil lineage¹⁵⁰.

Likewise, Mastrella et al. investigated the contribution of tumor cell-derived Apelin on tumor angiogenesis and discovered that the apelinergic system has a dichotomous role in angiogenesis and invasion in glioblastoma tumor models²⁵⁸. The mentioned study also emphasized the importance of F13A molecule, questioning its proposed antagonistic effect to APJ. Similar to therapeutic setup in this project, the authors proved that co-administration of DC101 and F13A was able to synergistically blunt both vascularization and invasive properties of the tumor.

However, complete diminishing of Apelin by terms of knocking down tumor-derived (tumor cell APLN-KD) and knocking out host-derived APLN (APLN-KO mice), tumor invasiveness increased. It was hypothesized that Apelin and F13A produced similar effect on APJ in terms of receptor-binding, however, downstream pathways might have been different depending on a ligand. Possible explanation that authors provide for this phenomenon is that F13A may not be able to sufficiently activate APJ in endothelium, but has a sufficient effect on tumor cells to stop the invasion and increase the overall survival²⁵⁸.

During this study, it became clear that this system requires optimization due to constant and random changes in metastatic burden, regardless of the experimental system’s invariability.

Addressing this issue presents a challenge and this model evidently requires a reporter that proves efficient cell engraftment. Moreover, re-evaluation of AAV vector’s functionality is necessary to ensure sufficient protein quantities in the targeted tissue. Finally, this experimental system needs to consider the usage of transgenic mice in terms of Apelin or even APJ depletion

DISCUSSION

130

(KO mice). It is worth noticing that usage of APJ-KO mice would disclose a valuable information on in vivo role of APJ in immunotherapy. On the other hand, it would be valuable to reveal the difference between the effects of cell- and host-derived APLN/APJ on metastasis formation in antiangiogenic conditions. However, in addition to the advantages from usage of Apelin-KO mice, it is worth to mention that such mice are completely Apelin-deficient which might result in undesired phenotype as physiological Apelin lacks in all tissues. The model used in this study still has a great potential and advantage only to target the lung ECs and should, therefore, be relatively specific for lung metastases – given that it would have worked and Apelin would have been overexpressed or antagonized sufficiently (F13A). Therefore, the improvement of this model would be of great interest for future work.

Few days before the conclusion of this thesis, the importance of the apelinergic system in contribution to tumor angiogenesis has been recognized in the study of Wang et al.²⁵⁵. Authors have successfully engineered Apelin-based synthetic Notch receptors (AsNRs) that can specifically interact with APJ and further stimulate synNotch pathways. That way, immune cells, engineered to express such receptor, were able to specifically target active tumor endothelium. The authors once more confirmed that apelinergic system is one of the key factors and a druggable target in inhibiting tumor angiogenesis. The potential of this system emerges slowly and may shed some necessary light in the area of such complexity as tumor angiogenesis.

DISCUSSION

131

Figure 47. Graphical representation of the results

CONCLUSION

132